Set up the Redis object cache with ChatGPT

Setting up a Redis object cache with ChatGPT means asking it in a chat to check a site's object cache, test the Redis connection and enable it through an xSpeed Hub app.

get_object_cache_statustest_object_cacheget_settingsupdate_settingstoggle_object_cacheflush_object_cache

Last updated: October 2026

Example session · ChatGPT

You: With xSpeed Hub, set up Redis object cache for shop and test it first.

get_object_cache_status(site: shop) read-only
→ Drop-in not installed. Backend: Redis configured.

test_object_cache(site: shop) read-only
→ Failed. Redis refused the login: wrong password.

ChatGPT: Redis is up but rejects the password the site has. Nothing is enabled. Update the password on the host or in wp-config.php, then tell me.

You: Fixed the password. Try again.

test_object_cache(site: shop) read-only
→ Connected. Probe value written and read back.

ChatGPT: The test passes now. Enable the object cache on shop?

You: Yes.

toggle_object_cache(site: shop, enabled: true) changes site
→ Drop-in installed. Redis is the object cache.

ChatGPT: Redis is on for shop.

How ChatGPT handles this

You are in ChatGPT with xSpeed Hub added as an app in developer mode, and you want a WordPress site to keep its database query results in Redis. Type: set up Redis object cache for the shop site and test it first. ChatGPT calls get_object_cache_status and reports whether the object-cache.php drop-in is installed and which backend is configured. It does not show whether Redis answers, so the status alone is never a go signal.

The next call is test_object_cache, which makes the site's plugin connect to Redis and read and write a probe value. Failures come back with the backend's own error: a wrong password, a refused connection or permission denied. Fix that on the host side, ask for the test again, and only then ask for toggle_object_cache with enabled set to true. ChatGPT asks you to confirm write actions by default, and because the Hub's tools carry no read-only annotation, it treats every Hub tool as a write, so expect a confirmation on get_object_cache_status and test_object_cache too. Expand a call to read what it will send. The Hub's instructions also tell the agent to briefly confirm intent for the object-cache toggles, but that is guidance to the model, not a lock.

What differs here is that ChatGPT is a chat app: no shell, so no redis-cli and no editing of wp-config.php. When the toggle runs, the site writes the connection constants into wp-config.php itself, so you rarely need the snippet; it can contain the Redis password in clear text, so keep it out of the chat. Write actions also depend on your plan: OpenAI's help center describes full MCP with write actions as a beta for Business, Enterprise and Edu, and says Pro accounts connect with read and fetch permissions. Custom MCP apps work on the web only, not in the mobile apps.

Set up ChatGPT once

Already connected? Skip to the prompts. Alternatives and troubleshooting are on the ChatGPT guide.

1Put your sites in xSpeed Hub

Sign in at app.xspeedcache.com with Google or email; the Hub is free and has no site cap. Then connect each WordPress site from its own dashboard: click Connect Hub in the xSpeed Cache top bar, then Connect via xSpeed Hub. Each site needs the free xSpeed Cache plugin.

2Turn on developer mode

In ChatGPT on the web, open Settings, then Security and login, and turn on Developer mode. On a Business workspace only admins and owners can use it. On Enterprise and Edu an admin grants access first, and you then turn it on under Settings, Apps, Advanced settings.

Text

Settings  >  Security and login  >  Developer mode

3Create the app from the Hub URL

Open ChatGPT Plugins, select the plus button and create a developer-mode app. Name it xspeedhub, paste the MCP server URL and choose OAuth. Click Scan Tools, approve access on the xSpeed Hub page that opens, wait for the scan to list the tools, then click Create. Workspace admins can also start from Workspace settings, Apps, Create.

Text

https://app.xspeedcache.com/xspeed/mcp

4Select the app in a chat

The app appears under Drafts in your app settings with a Dev label. In a new chat, open the plus menu in the composer, choose Developer mode and select xspeedhub. In a workspace, only an admin or owner can publish the app for other members.

Text

+  >  Developer mode  >  xspeedhub

1 / 4

Full ChatGPT setup, sign-in options and FAQ

Before you send a prompt that changes something

Reads change nothing on your sites, though contact_support emails xSpeed support. Writes run once your connection allows them, so your client's approval prompt and a read-only connection (the token's Read-only everywhere switch, or a Viewer sign-in) are the gates that matter. ChatGPT asks you to confirm write actions by default. You can expand a tool call to read the JSON it will send, and you can choose to remember an approve or deny choice for a tool for the rest of that conversation. ChatGPT respects a tool's read-only annotation and treats a tool without one as a write action; xSpeed Hub's tool definitions do not declare that annotation today, so expect confirmations on reads as well. OpenAI says confirmation also depends on app permissions and context, and some especially risky actions can be blocked. xSpeed Hub has no confirmation step of its own, so once you tell ChatGPT to remember Approve for a tool, that tool runs without a prompt for the rest of the conversation, as soon as your connection allows writes.

What do I ask?

Three prompts written for ChatGPT. More for this job are below.

Prompt for ChatGPT

Using xSpeed Hub, check the object cache status on shop and tell me whether Redis is configured.

Prompt for ChatGPT

Use xSpeed Hub to run the Redis connection test on shop. If it fails, explain the error and what to change on the host.

Prompt for ChatGPT

Ask xSpeed Hub what is configured for Redis on shop, without showing me any password.

What happens, step by step

A persistent object cache keeps database query results in Redis or Memcached instead of rebuilding them on every request. Through xSpeed Hub an agent can read the state of the cache, probe the cache server, switch the cache on or off, and flush it. The Hub tells the agent to run the probe first, because the status call alone does not prove the server is reachable.

Read the current state

get_object_cache_status reports whether the object-cache.php drop-in is installed, which backend is configured, Redis or Memcached, and its stats. It does not prove the server answers, so a clean status is not a green light.

Test the connection first

test_object_cache opens a live connection to the configured server and does a read and write probe with the settings saved on the site. A failure comes back with the server's own reason, such as a wrong password, a refused connection or a permission error. The Hub tells the agent to run this before toggle_object_cache.

Fix the connection if the test fails

get_settings reads the object-cache module, with secrets masked. update_settings changes it, for example the host or port. On a managed host that defines these values in wp-config.php, the fields are read-only and a write that tries to change one is refused with the name of the constant. For a host that wants the constants pasted, get_object_cache_status with detail set to snippet returns the wp-config.php block.

Turn it on

toggle_object_cache with enabled true makes the site test the connection again, write the connection constants into wp-config.php and install the object-cache.php drop-in. If the test fails, it stops with "Could not enable" and writes nothing. If wp-config.php is not writable, the drop-in goes in and the reply tells you to add the snippet by hand. The Hub tells the agent to briefly confirm intent first.

Check it took

The drop-in loads on the next request, so the agent reads get_object_cache_status again after a page load and runs test_object_cache once more. The status also warns when the cache connects but refuses writes, so a connected cache that stores nothing does not pass as healthy.

Flush when stale data survives

flush_object_cache empties the persistent object cache. Use it when old options, transients or query results outlive a page cache purge. It does not touch the page cache; purge_cache does that.

Reference

Read tools	get_object_cache_status and test_object_cache, both work on a read-only connection
Write tools	toggle_object_cache (enabled true or false) and flush_object_cache
Backends	Redis or Memcached
Status detail	status (default) or snippet, the wp-config.php constants block
What status proves	Drop-in installed, backend configured, stats. Not that the server is reachable
What the test proves	A live connect and a read and write probe on the saved settings, with the server's own error on failure
Enable does	Tests the connection, writes connection constants to wp-config.php, installs the object-cache.php drop-in
Disable does	Removes xSpeed's drop-in and its wp-config.php block; a drop-in owned by another plugin is left alone
Connection timeout	1 second by default, set in the object-cache module
Settings module	object-cache, for get_settings and update_settings
Every site at once	None of these tools accept site: "all"; one site per call
Confirmation	None on the Hub; your client's prompt is the gate

Rules worth keeping

Test before you toggle. The Hub tells the agent to run test_object_cache first, because a cache server that does not answer slows every request while the page cache still looks healthy.
Keep the Redis password out of the chat. get_settings masks it, but the wp-config.php snippet is built from the saved settings and can include a saved password. Set secrets in wp-config.php or in the xSpeed panel.
toggle_object_cache and flush_object_cache are write tools. The Hub tells the agent to briefly confirm intent for the object-cache toggles, but that is guidance, not a lock. Your client's approval prompt and a read-only grant are the gates that matter. A read-only grant means the connection token with Read-only everywhere on, or a Viewer member's sign-in; an OAuth sign-in gets the scopes the client asks for, which are read and write by default.
If another plugin owns the object-cache.php drop-in, turn its object cache off in that plugin. xSpeed leaves a drop-in it did not install in place when you disable.
A flush clears cached data, so the next requests rebuild it from the database and the site is briefly slower. Flush one site, then check it, before you repeat it.

Good to know with ChatGPT

If ChatGPT refuses the toggle, check your plan before you suspect Redis: Pro accounts connect with read and fetch permissions. The status and test calls can still work, so you can finish the diagnosis in chat and switch the setting on from the Object Cache panel in wp-admin instead. Also watch the remember option: once you tell ChatGPT to remember Approve for toggle_object_cache or flush_object_cache, that tool runs without a prompt for the rest of the conversation, and the Hub adds no confirmation of its own.

More prompts for this job

They work in any client connected to xSpeed Hub.

Prompt

Using xSpeed Hub, is a persistent object cache running on shop, and which backend does it use?

Prompt

Use xSpeed Hub to test whether Redis is reachable from shop. Do not change anything.

Prompt

Shop is slow in wp-admin. Use xSpeed Hub to check the object cache before anything else.

Prompt

With xSpeed Hub, test Redis on shop, and if it passes, turn the object cache on.

Prompt

Ask xSpeed Hub for the wp-config.php object cache snippet for docs.

Prompt

Use xSpeed Hub to flush the object cache on blog. Old options survive a page cache purge.

Prompt

With xSpeed Hub, turn the object cache off on shop and confirm the drop-in is gone.

Frequently asked questions

Yes, if your plan lets a custom app run write actions, which OpenAI describes as a beta for Business, Enterprise and Edu. It calls test_object_cache first, then toggle_object_cache. If your plan limits writes, ChatGPT can still read the status and run the test, and you switch Redis on in the Object Cache panel in wp-admin.

It reports the error from the Redis server, such as a wrong password or a refused connection, and should stop before it enables anything. The site also refuses to enable when its own connection test fails, so a failed test means fixing the host side and testing again.

Because a cache server that does not answer costs time on every request while the page cache still looks fine. The Hub tells the agent to run test_object_cache first, and the site also tests the connection when you enable and refuses to continue if it fails. The probe is still worth running on its own, since it returns the server's exact reason, such as a wrong password or a refused connection.

The site keeps working. The drop-in is built never to take the site down: it falls back to WordPress's own cache for the length of the request and marks itself as degraded. A request can still wait up to the connection timeout, which is 1 second by default, so the site gets slower. xSpeed reports that state as degraded, and test_object_cache names the fault.

No. These tools connect xSpeed to a Redis or Memcached server that already exists and switch the object cache on or off. Getting a server is a hosting task. xSpeed uses the phpredis extension when it is there and its own built-in client when it is not.

flush_object_cache empties the persistent object cache and nothing else. purge_cache with type object does the same job through the purge tool, and purge_cache with type all also clears the page cache and, when connected, the Cloudflare edge. Use the flush when stale options or transients survive a page cache purge.

Treat it like a secret. The snippet is built from the settings saved on the site, so it can contain the Redis password if one is saved. Paste it into wp-config.php yourself and avoid sharing it in a chat or a ticket.

Set up Redis object cache, from ChatGPT.

xSpeed Hub is free and has no site cap. Connect your sites once and ask from the agent you already use.

Open xSpeed Hub ChatGPT setup

Set up the Redis object cache with ChatGPT

How ChatGPT handles this

Set up ChatGPT once

What do I ask?

What happens, step by step

Reference

Rules worth keeping

More prompts for this job

Frequently asked questions

Keep going

Set up Redis object cache, from ChatGPT.